Weather variability presents an external impediment that increases the uncertainty in crop production potential specially for perennial crops under rain-fed cultivation. In particular, increasing rainfall variability that has accompany climate change in the tropics is increasing production uncertainty exposing farmers to a range of risks. Producer behavior under uncertainty is theoretically advanced in Chambers and Quiggin (2000) with the development of the state-contingent approach. This theory advocates farmers manage uncertainty by allocating inputs into different states of nature. This will enhance farmers' capacity to choose variable production technologies to manage uncertainty associated with unexpected events as state-contingent production technologies are naturally more flexible. This thesis examines the application of a state-contingent model for coconut production to assess the impact of a changing climate on coconut production with a view to assess adaptation options for a production system involving long-lived crops. The analysis of production of perennial crops encounters the treatment of long gestation periods, gradual increase in yield over an extended maturation period and yield stabilization following maturity. In considering optimal resource allocation from an agricultural production system perspective, a) the nature of resource constraints, b) the appropriateness of production technology and c) decision makers managerial behaviour needs to be considered together. The analysis in this thesis integrates these three aspects, with a special focus on managing uncertainty in coconut production in Sri Lanka. Climate is a primary determinant of crop choice and weather variability is the major factor influencing agricultural productivity and uncertainty in production outcomes. To examine the impact of changes in climatological variables, rainfall and temperature, on coconut production, a production time series over 1950-2014 was used to identify variations and trends over time and their statistical significance. The graphical and statistical analyses undertaken proved useful in gaining a more complete understanding of how the production environment is changing in the short run as well as over the long run. It offered a sound basis to further consider effective adaptation options in agricultural production, especially long-lived assets such as coconuts, in response to changing states of climate, representing the natural environment in which production takes place. It was revealed that while climate change impacts are observable over longer time horizons, understanding the nature of such changes may allow farmers to adapt to more observable variations in weather over narrow time periods.